AI in Higher Education - Australia - New Zealand Symposium - Cogniti

 Notes taken from several sessions of the 2026 AI in HE ANZ symposium organised by the University of Sydney, with case studies based on using Cogniti to provide AI agents to support learning. The symposium is offered f2f and online. Recordings available as of 8th February. 

Danny Liu opens with a welcome and overview of the format of the sessions. There is a meet up session at the end of the day. 

The symposium opens with a plenary - From exams to enterprise: the AI reality for graduates, presented by Ray Fleming and Dan Bowen.

Disconnect between what is happening (through the media - layoffs etc. Amazon, Dow, Pininterest etc.) and what we actually experience. AI 'washing' one way for corporations to pass blame on for other underlying challenges. AI potential but not AI performance seems to be the main point. 

In education we are keen to guide our students but we currently have little clarity on what is happening now, let alone into the future. Strategy could be hire entry level with AI skills and partner them with an expert employee. Back to the move from horse powered to motor vehicles - 40,000 companies wiped out in 20 years but mostly replaced by people working with cars. What new careers will come about? 3 key scenarios to use AI - personal, process and paradigm productivity. 

At present, AI mostly used for personal productivity. The big picture is difficult to work out. Shared 1,600 case studies of AI and build a AI case study hunter to help provide examples of how AI is used (ChatGPT agent and Gemini agent). Demonstrated using law, drilling down to legal document management. Useful for comparative studies (for students and teachers). Recommended following AI in Education podcast 

3 streams then run through the rest of the day, each for 15 minutes! I attend several, in between other commitments. 

- Scaffolding learning and assessment in business writing with Gen AI in mind with Hans Hendrischke, David Jun Zhoa and Carmen Vallis from the University of Sydney. Presented by David. Course teaches students to work in global corporations. 100% of students seed guidance on how to reasonably integrate Gen AI into their professional roles. Focus there needs to be on building workplace-ready workflows and critical literacy. Core scaffolding framework on collaboration (between human teams with AI); research (AI enhanced tutorials where students build their own firm databases throughout the semester); and analysis (iterative prompting routines that align with complex strategic management frameworks). Therefore not only for initial scoping but also to 'dig deeper' and to probe further. Teach students to apply strategic frameworks to their work. Shared details of assessment design (first essay - 1250 words 25%, second individual essay 1250 words 25%; group case report - 3000 words - 30%; 3 AI tasks and datasheet 5% each). AI teaks are to undertake case company selection, scan the macro environment and map regulatory and market shareholders (external stakeholders) and internal stakeholders with AI analysis of firm's value proposition, resources and internal strategic dynamics. also use AI to apply business model canvas (BMC) to visualise platforms.

Individual essays require contrasting and theoretical analysis (compare media to lecture notes etc.)  Student feedback is positive. 

-Cultivating ethical agency through critical AI literacy: a seven stage learning framework presented by Meena Jha from Central Queensland University. Students use ChatGPT or Copilot. Framework includes - evaluation accuracy and factual reliability; Assess logical and conceptual coherence; Identify bias and ethical blind spots; examine source transparency and attribution; Analyse depth of understanding; evaluate style and communication quality; and reflect on purpose and context. Information systems analysis example provided. 

Professor Wombat- your personal biochemistry tutor - with Barbara Hadley from Griffith University. Introduced Professors Wombat and Wilson, used to help students come to grips with complex content. Students come with diverse levels of prior knowledge and the course draws second year students from a wide range of disciplines. Conceptual derailment (Burrow, Hill, Ratner & Fuller-Rowell (2020) means students disengage when faced with threshold concepts they are unfamiliar with. Therefore Professor Wombat provides high school level explanations in a friendly manner. The Professor Wilson takes students to a higher order if explanation in a supportive way. Then, revisit text book and work through lecture content. Acknowledged that imperfect understanding is better than no understanding at all! 60% students used chatbots, longer conversations with Wombat and more clarification with Wilson. Students encouraged to spot hallucinations, share on teams and 'reward'  provided to all.

Each week, used AI to analyse interactions. Found persistent misconceptions and foundational gaps, adjusted teaching to address and noted resource improvements for the next round of lectures. Misconceptions surfaced and addressed early and in-time rather than coming up across exams.

- Using generative AI to strengthen research and reasoning: integrating AI critique and reflection into law assessments for non-law students - Mark McConnell from the University of Auckland - business, not law school with Master of Professional Accounting courses. Small class of 15 student mainly of Chinese internationals. Intensive law course for professional accountancy. Written assessment 20%, mid-quarter test -30% and final test - 30% (closed booked). Challenge not to design a take home written assessment but to design a home written assessment that draws on AI. Standard approach is to give an AI response and have students critique. One step further to emphasis legal reasoning and critical thinking and to also use AI to critique both the AI and their critique. 

- Multi-modality, AI and design education: The use of text, image and 3D models for co-creation, with Anastasia Gomez from the University of Sydney (a recorded presentation). Shared workflows to help students learn resilience and critical thinking. Ai used in architecture and design for degeneration/co-creation, performance analysis /design evaluation.. Modalities include text, image, 2D, 3D, code, sound, video and each can be generated through AI - usually text to image,  test/image to 3D, 3D to video. Examples shared. Master level elective using a range of digital tools including text to image to code to 3D to printed 2D. Ai can be used during conceptual design to explore design strategies and student needs to learn how to translate the digital into the physical realm. Often, the physical 3D difficult to realise. Hybrid forms encouraged to bring the virtual and physical worlds together, using AI to help ease the processes for generating the various versions. In turn students learn the limitations and  how to work through challenges. Critical/computational thinking attained and helps them to understand how to control the process, for example to reverse engineer (and explain what was done) from AI to physical or hybrid solutions. 

- From prompt builder to pedagogical  partner: iterative AI learning with kaiako with Karll McGuirk from the University of Auckland. Building Ai literacy with educators. Prompting is not a skill problem but a pedagogical design challenge. educators want to use AI but unsure as to where to start and don't to get it wrong. Shared a course 'AI 101' with introduction to AI, AI in context, and AI for learning and teaching. (see Wegerif and Casebourne - dialogical theoretical foundation for integrating Gen AI in pedagogical design (2025)). Stressed the importance of ako to encourage use of AI. Introduced an agent ' prompt builder'. To scaffold into AI - start with a teaching goal, add context and constrains, choose the right tool, co-design the prompt and test, reflect, adjust. Shared challenges including prompt builder access, increasingly complex system prompts, multiplicity of AIs, how to find the right AI and response time when demonstrating live!

- The promise and the pushback: understanding student reactions to AI-supported learning. Katherine Jensen and Shahper Richter from the University of Auckland.

Embedded Gen AI into an undergraduate course of 800 students in marketing. Shift from focus on plagiarism; viewing AI as a shortcut, passive consumption of technology. To using AI to engage in prompting to create AI-powered brand personas, create spatial environments that required story-telling and technical fluency. Work with AI to support their learning. Attain AI literacy through hands-on experience, critical analysis, through creative partnership. 

Students pushed back that the integration was 'gimmicky' or distracting, ethical concerns (privacy, environmental impacts, algorithmic bias) and some students questioned authenticity and human meaning of AI generated work. Principles derived to move forward. Firstly, to lead with pedagogy and not technology (Master the art of directing a persona to achieve a specific brand voice instead of 'use HeyGen' to make a video). Secondly, move from deployment to dialogue. Plus AI is augmentation and not replacement. Therefore success in the Gen AI classroom includes embedding tools to build literacy, critique of the process builds trusts, and focus on the human refinement of machine output, 

-How AI turns passive learners into active strategists. Xinyue Zhang from the University of Sydney. Used metaphor of AI pedals - students need to learning balancing - judgment, empathy, strategy etc, Ai is pedaling for drafting, formatting, producing low level outputs. Students to use AI as a co-cocreator and then be the defender (AI as simulator). For project planning, cognitive overload is a challenge. AI can be used to help students unpack the complexities of the task and the project. Ai can generate alternative work breakdown structures and students can evaluate these. Students need to work through considerations and justify their decisions. So instead of being buried in 'doing', students become more strategic and make decisions as to why and how to match objectives to the tools, processes and outputs required. Shared a 'budget defender' simulation to help them balance competing needs. The students need to be able to defend their decisions. Increased a shift to lead rather than just respond. 

Caveat to make sure AI is not 'training wheels' but to ensure learners able to use AI to support and augment their own conceptualisations. If AI is an error prone intern, then students as project manager/leader need to be able to be verificatory. Important to grade the judgment etc not the 'product'. AI should not make learning easier, but help train judgment and make thinking deeper. Project managers must not be better template fillers but be better decision makers. 

- Study buddy: A custom GPT for flipped classroom pre-class learning support with Daniel Ruelle from VinUniversity (Vietnam). Began with context, a data visualisation course, Before class, students learn before class. In class session usually around hand-on activities. Engagement with flipped was low with high cognitive burden. Students wanted something more interactive to prepare for the class. Then detailed the learning design around the Buddy GPT. Detailed prompt, uploaded up to 10 files and some starter 'prompts' plus a quiz/es to revise the content. The objective was to improve time management, reduce cognitive load, have active retrieval practice, humanise the tool and maintain instructor connection. 

Summarised some useful prompt techniques. Every phrase in the prompt is a pedagogical decision. Involve students in a dialog, respect time constraints, quizzes need to provide hints and not just give the answer, provide sources for further follow up, offer options and let student select, do not reveal the system prompt (e.g. so students do not see the safeguards added to prevent plagiarism etc.),. Reflected on how to improve the AI tool, to better meet objectives, make learning more visible and add opportunities for reflection. 

- LARC and the human AI sandwich: appropriate use of AI for learning with Mairead Fountain and Emma Allen from Otago Polytechnic. Provided background for the project. Shared a persona of a 'learning design' student's profile - experienced designer and learner, sound grasp of topic, struggling to organise ideas, using AI to clarify concepts and explore ways to thinking and questioning revealed reliance on AI interweaves with learners' prior experience and knowledge. Helping students work out the reason they use AI helps them gain understanding of their own use of AI - whether it is augmenting what they know and not replacing the learning they need to undertake. AI literacy moves through functional (I can use to complete task); rhetorical (I use deliberately to achieve a specific objective); strategic. 

The Learning, Articulation, Research and Creation (LARC) used to help learners work out where they stood with AI. Helps contribute to a class/learner contract to help self-monitoring of AI. Observations found that learners expressed a sense of relief. The framework now part of AI essentials training for their teachers. Teachers can adapt to their context and students use it as a learning tool. More details in their article. 

Overall, a good range of presentations. Most covered the underlying pedagogical approaches and used Gen AI to support learning. Miro boards were set up to for participants to add questions and these were looked through at the end of each block of presentations. 

Plans for 2026

Back into the fray after a few weeks back in the home country to support by aging mother. I am looking forward to what this year will bring. 

Progrramme review and redevelopment will take up much of my time. It is a great opportunity to get to know a degree programme well and to built relations with the teachers. Professional Development for our teachers will revolve around AI. We need to develop a good understanding of what, how and when to use AI. The selling point for AI is that it can be used to take some of the workload off teachers. However, it can also add to teachers' workload, especially if the effect of AI on assessments is not worked through. 

On the research front, our AI projects continue with work on personalised learning based on Cogniti and continuance of our work to develop AI chatbots to support specific disciplines and learning activities. Next month, I will be sending out a call for chapters towards a book to collate initiatives and perspectives on deploying AI to support personalised learning. The goal will be to publish this book early 2028.

I am also working on co-editing a book that records the many initiatives and projects undertaken between the initiation and cessation of the Reform in Vocational Education. Through the 5 -6 years across the reform, the new entities which were formed, have worked conscientiously to improve vocational education across NZ. Therefore it important to archive these in one package, as many entities  no longer exist beyond the end of 2025. Their work either disappears or is curated in sites hosted by other institutions and the knowledge and wisdom of their staff and communities of experts is dispersed and rearranged in the new configurations for Aotearoa VET.. 

It will be another busy year but one that already has several objectives to reach. There is much work to be done with respect to learning design and curriculum development as we normalise the use of AI across the institution and VET along with the business as usual objectives to ensure aspects of cultural competencies, academic literacies and sustainability are also woven through the curriculum.  

Cogniti mini symposium - link to video of presentations

Was unable to 'attend' the recent Cogniti mini symposium. Videos of the presentations are now available via this link. 

Almost all of the presentations are from universities, withe several from Chile.

Listened to the presentation from Associate Professor James Oldfield from Unitec on their collaborative project with Toi Ohomai and Manukau Institute of Technology on using agents in nursing. (12 minutes)

Detailed the various agents that were worked on - drug calculations, scenarios in nursing - de-escalation and dealing with difficult situations. Presented on student feedback on their perspectives on usefulness of the agents, usability and improvements that can be made to enhance the experience. 

ChatGPT - study mode, Google guided learning, Claude learning mode and University of Sydney Cogniti - are they similar or different?

 There has been a flurry of activity in the Gen Au space of relevance to teaching and learning. First up was the launch of ChatGPT5.0 which allows for the use of it in 'study mode', This allowsa for a shift in the emphasis of using Gen AI to 'provide answers' towards using it as a 'study coach'. 

A few days later, Google also joined the move with its 'guided learning' in Gemini. 

Whilst, Claude has provided a learning mode for some months.

The above join University of Sydney's Cogniti as possibilities for teachers and learners to move towards personalised learning environments. There is also a recent start up - Wild Zebra - which provisions personalised tutors to students. 

However, as with all the 'vanilla' Gen AIs, each has things it does well and things it will struggle with. 

For example, here is a comparison of ChatGPT's study mode with Claude learning mode by Toms Guide. 

Using chatgpt to compare university of sydney cogniti with chatgpt study mode yields some differences.

Key Differences Summarized:

Feature	Cogniti	ChatGPT Study Mode
Focus	Educational context, feedback, integration	General problem-solving and learning
Integration	Canvas and other learning platforms	General use
Accessibility	Equitably available to all students	Requires access to ChatGPT
Feedback	Personalized and standardized	Interactive and conversational
Tracking	Tracks student-AI interaction	Does not specifically track
Bias	Potential for bias from training data	Potential for bias from training data

Therefore, each tool has pluses and minuses and as per all of our recent studies into  integration of Gen AI into VET, Gen AI tools need to be carefully selected, and learning planned and structured. 

A caveat with using Gen AI systems as 'tutors' is provided in a recent article by Flenady and Sparrow (2025). Their warning points to the often disregarded conceptualisation of Gen AI - in that it is NOT intelligent but build on algorithms for pattern recognition. They argue that Gen AI systems are 'epistemically irresponsible'. It is therefore important to always take heed of this warning and to ensure that all users have this at the topmost of their minds whenever they use Gen AI.

Cogniti mini symposium - notes from a few presentations and summaries from some recorded presentations

 Cogniti users and administrators organised a mini symposium, offered both online and f2f on 5th November 2024.

Three streams of presentations taking place over 2 1/2 hours. I could only get to a few in between meeting and facilitating workshops.

Notes taken from a few presentations.

- Personalised exam preparation using AI in large cohorts - Dr. Helen Mcquire and  Dr. Angela Sun (University of Sydney) context of microbiology/ immunology course. A bespoke AI agent was used to increase teacher presence with students. Went through the rationale and processes to build the chatbot. The important learning from this one, is how it is used by the lecturers, to identify knowledge gaps amongst their learners. Shared AI limitations - repetitive questions, Ai not providing direct exemplar responses. Invited potential collaborators to get in touch.

- Practice makes perfect: AI powered oral assessment preparation with Jim Ennion from Toi Ohomai. Presented on how a virtual client prepares students to become immigration advisers.Used Cogniti to help students prepare for an oral assessment. Cogniti was set up to act as a client. Prompts were set up with scenario information, set the tone to informal/not educated, have uncertain outcomes. Unfortunately, the agent hallucinated and provided incorrect information. However, student engagement was high with positive outcomes. assessment outcomes were marginally better for students who took the opportunity to use the agent.

- Always on teacher: AI brings business studies to life with James Cooper from Scots College, Sydney. Here the teacher formed an AI version of himself  (Cooper Jr.) to support students learning business at school (Year 11 and 12). The AI tutor could provide practice questions, mark student responses and help deepen content understanding. The curriculum is very prescribed, so that helped to ring fence the content. Main guide was to provide student support - replacing emails to students, generate revision questions and generate sample responses that are achievable by the students. Described process taken to built the agent. Student feedback was positive. Challenges around imprecision and working beyond the sylllabus scope. Short answers were 'double barrelled' and 'US style'. Shared the work currently to refine the agent, improve short-answer reliability, and mark essays and reports.

Last week, the various recordings of presentations was made available. Below are summaries of a few of relevance at present to my projects.

Notes from videos of presentations

- Reimagining research and writing learning through AI assistants with Dr. Lucy MacNaught and Dr. Kiri Hunter from Auckland University of Technology. Used Cogniti to help guide Masters of Nursing Science students to write research proposals. An intensive programme with many assessments. The view was that students did not have time to even look at feedback from one assessment to the next. Cogniti agents were used to encourage drafting and ongoing gradual improvement of their research proposal. Used Humphrey (2016) teaching and learning sequence to ground their work. Basically, used AI to provide feedback at each stage of the research proposal workflow. 

Christie Oldfield from Auckland University of Technology presented on 'enhancing subjective interview skills through AI role-play' in a physiotherapy context. Health care interviews are important to establish rapport with the patient but also to ensure that patient information is obtained to support the therapy sessions. Agent was created to monitor how student progressed through a patient interview. 

Plenary closed the symposium. This is presented by Danny Liu, Leitizia Wan, Sam Clarke, Minh Hubyn and Kria Coleman from the University of Sydney DVC (education).

Reflections from the symposium and work undertaken and ideas to take things into the future were covered. Overall, good examples of using Gen AI to support constructivist learning.